3 ms·
The main “cheat” that occurred is AlphaStar’s definition of APM. They capped APM averaged over an entire game. Essentially it saved its actions it didn’t need
by jnwatson 8y ago
The main “cheat” that occurred is AlphaStar’s definition of APM. They capped APM averaged over an entire game. Essentially it saved its actions it didn’t need at the beginning of the game and unleashed a torrent of superhuman >1000 APM for a solid 30 seconds in that stalker battle. That should be accounted for in the next demo.
- RoboTeddy 8y agoThe averaging period doesn't last the whole game: "we set a maximum of 600 APMs over 5 second periods, 400 over 15 second periods, 320 over 30 second periods, and 300 over 60 second period. If the agent issues more actions in such periods, we drop / ignore the actions. These were values taken from human statistics." [1] I'd be curious about the approach they used to draw from human statistics. It's possible that peaks in the human APM distribution are caused by e.g. holding down keys, resulting in ~2000 APM [2]. [1] https://www.reddit.com/r/MachineLearning/comments/ajgzoc/we_are_oriol_vinyals_and_david_silver_from/eexs0pd/ https://www.reddit.com/r/MachineLearning/comments/ajgzoc/we_... [2] https://blog.usejournal.com/an-analysis-on-how-deepminds-starcraft-2-ai-s-superhuman-speed-could-be-a-band-aid-fix-for-the-1702fb8344d6 https://blog.usejournal.com/an-analysis-on-how-deepminds-sta...
- setr 8y agoIts also very unlikely that an AI click is equivalent to a human click — humans will spamclick to ensure the action goes through, while the machine can do it once with full confidence